To run the pipeline you need to do some setup. The exact steps you need to take depends on the platform you are running the pipeline on, and will be detailed below and in HOWTO. Independent of platform, running the pipeline is done using caper and (optional but recommended) output organization is done using croo. Both caper and croo require python version 3.6 or newer.
Direct usage of the execution engine Cromwell features complicated backend configuration, workflow options and command line parameters. Caper hides the complexity and consolidates configuration in one file. Caper is available in PyPI and it is installed by running:
$ pip install caperCaper version 0.8.2.1 or newer is required.
Note that conda run mode that is described in caper documentation is not supported by this pipeline.
The way Cromwell organizes pipeline outputs is not always the most clear and findable. Croo is a tool to reorganize the files in more readable manner. Croo is available in PyPI and it is installed by running:
$ pip install crooJava is required to run execution engine Cromwell that caper uses under the hood.
To check which Java version you already have, run:
$ java -versionYou are looking for 1.8 or higher. If the requirement is not fulfilled follow installation instructions for mac or linux or use your favorite installation method.
Pipeline code is packaged and distributed in Docker containers, and thus Docker installation is needed. Follow instructions for mac or linux.
If you want to use Singularity instead of Docker, install singularity. Pipeline requires singularity version >=2.5.2, the link takes you to version 3.1.
If you are intending to run the pipeline on Google Cloud platform, follow the caper setup instructions for GCP.
- For an example on how to run the pipeline on Google Cloud, see HOWTO.
If you are intending to run the pipeline on AWS, follow the caper setup instructions for AWS.
Behind the scenes, caper uses Cromwell. Pipeline has been tested using version 49 of Cromwell.